Technical Indicator Optimization for Indian Stocks

Introduction

Technical Indicator Optimization represents a paradigm shift from traditional return-based portfolio optimization to signal-based optimization. Instead of using historical returns and covariance matrices, this approach leverages cross-sectional z-scores of technical indicators to construct portfolios that capitalize on relative strength and momentum patterns across assets.

This methodology transforms the portfolio optimization problem from a quadratic programming problem (as in Mean-Variance Optimization) into a Linear Programming (LP) problem, making it computationally efficient and robust to estimation errors inherent in return forecasting.

The approach is particularly well-suited for the Indian equity markets (NSE/BSE) where technical analysis has historically shown strong predictive power, and where the cross-sectional dispersion of technical signals can be effectively captured and monetized.

Intuitive Explanation

Imagine you're a cricket scout looking for the best players for your team. Instead of just looking at each player's individual statistics (like traditional portfolio optimization looks at individual stock returns), you compare how each player performs relative to their peers across multiple skills: batting average, bowling speed, fielding accuracy, etc.

Technical Indicator Optimization works similarly. For each stock, we calculate multiple technical indicators (RSI, Moving Averages, Williams %R, etc.), then determine how each stock ranks relative to all other stocks for each indicator. A stock with consistently high relative rankings across multiple indicators gets a higher allocation in the optimized portfolio.

Example: If RELIANCE has an RSI of 70 when the average RSI across all Nifty stocks is 50 with a standard deviation of 15, then RELIANCE has a z-score of +1.33. If it also has positive z-scores for momentum and moving average indicators, it receives a higher weight in the optimized portfolio.

Technical Indicators: Mathematical Foundations

Our optimization framework incorporates 12 different technical indicators, each capturing different aspects of price momentum, trend, and market psychology. Here are the mathematical formulations:

1. Simple Moving Average (SMA)

The SMA smooths price data by calculating the arithmetic mean over a specified period:

SMA_n(t) = \frac{1}{n} \sum_{i=0}^{n-1} P_{t-i}

Where $P_t$ is the price at time $t$ and $n$ is the lookback period.

2. Exponential Moving Average (EMA)

EMA gives more weight to recent prices, making it more responsive to new information:

EMA_t = \alpha \cdot P_t + (1-\alpha) \cdot EMA_{t-1}

Where $\alpha = \frac{2}{n+1}$ is the smoothing factor.

3. Weighted Moving Average (WMA)

WMA assigns linearly decreasing weights to older prices:

WMA_n(t) = \frac{\sum_{i=0}^{n-1} (n-i) \cdot P_{t-i}}{\sum_{i=0}^{n-1} (n-i)}

4. Relative Strength Index (RSI)

RSI measures the magnitude of price changes to evaluate overbought/oversold conditions:

RSI = 100 - \frac{100}{1 + RS}

Where $RS = \frac{Average\ Gain}{Average\ Loss}$ over the specified period.

5. Williams %R

Williams %R compares the current close to the high-low range over a lookback period:

\%R = \frac{Highest\ High - Close}{Highest\ High - Lowest\ Low} \times (-100)

6. Commodity Channel Index (CCI)

CCI measures the variation of price from its statistical mean:

CCI = \frac{Typical\ Price - SMA(Typical\ Price)}{0.015 \times Mean\ Deviation}

Where $Typical\ Price = \frac{High + Low + Close}{3}$

7. Rate of Change (ROC)

ROC measures the percentage change in price over a specified period:

ROC_n = \frac{P_t - P_{t-n}}{P_{t-n}} \times 100

8. Average True Range (ATR)

ATR measures market volatility by decomposing the entire range of an asset price:

TR = \max(High - Low, |High - Close_{prev}|, |Low - Close_{prev}|)

ATR = SMA(TR, n)

9. SuperTrend

SuperTrend combines ATR with price to create dynamic support/resistance levels:

Basic\ Upper\ Band = \frac{High + Low}{2} + (Multiplier \times ATR)

Basic\ Lower\ Band = \frac{High + Low}{2} - (Multiplier \times ATR)

10. Bollinger Bands

Bollinger Bands use standard deviation to create dynamic bands around a moving average:

Upper\ Band = SMA + (k \times \sigma)

Lower\ Band = SMA - (k \times \sigma)

Where $k$ is typically 2 and $\sigma$ is the standard deviation.

11. On-Balance Volume (OBV)

OBV combines price and volume to show buying/selling pressure:

OBV_t = \begin{cases} OBV_{t-1} + Volume_t & \text{if } Close_t > Close_{t-1} \\ OBV_{t-1} - Volume_t & \text{if } Close_t < Close_{t-1} \\ OBV_{t-1} & \text{if } Close_t = Close_{t-1} \end{cases}

12. Accumulation/Distribution (A/D)

A/D line shows the relationship between price and volume flow:

Money\ Flow\ Multiplier = \frac{(Close - Low) - (High - Close)}{High - Low}

A/D = Previous\ A/D + (Money\ Flow\ Multiplier \times Volume)

Cross-Sectional Z-Score Methodology

The core innovation of Technical Indicator Optimization lies in transforming absolute indicator values into relative rankings through cross-sectional standardization.

Step 1: Calculate Technical Indicators

For each asset $i$ and each technical indicator $j$ , we calculate the indicator value $I_{i,j}(t)$ at time $t$ .

Step 2: Cross-Sectional Standardization

At each time period $t$ , we standardize each indicator across all assets:

Z_{i,j}(t) = \frac{I_{i,j}(t) - \mu_j(t)}{\sigma_j(t)}

Where:

$Z_{i,j}(t)$ : Z-score of asset $i$ for indicator $j$ at time $t$
$\mu_j(t) = \frac{1}{N} \sum_{i=1}^{N} I_{i,j}(t)$ : Cross-sectional mean of indicator $j$
$\sigma_j(t) = \sqrt{\frac{1}{N-1} \sum_{i=1}^{N} (I_{i,j}(t) - \mu_j(t))^2}$ : Cross-sectional standard deviation

Step 3: Signal Aggregation

We combine multiple z-scores to create a composite signal for each asset:

S_i(t) = \sum_{j=1}^{M} w_j \cdot Z_{i,j}(t)

Where $w_j$ are the weights assigned to each indicator (equal weighting by default: $w_j = \frac{1}{M}$ ).

Linear Programming Formulation

The technical indicator optimization problem is formulated as a Linear Programming problem that maximizes the expected portfolio signal while satisfying practical constraints.

Objective Function

We maximize the portfolio's expected signal strength:

\max_{w} \sum_{i=1}^{N} w_i \cdot S_i

Constraints

1. Budget Constraint

The portfolio weights must sum to 1 (fully invested):

\sum_{i=1}^{N} w_i = 1

2. Long-Only Constraint

No short-selling is allowed:

w_i \geq 0 \quad \forall i

3. Maximum Weight Constraint

To ensure diversification, we limit individual asset weights:

w_i \leq w_{max} \quad \forall i

Where $w_{max}$ is typically set to 30-40% to prevent concentration risk.

Complete Mathematical Formulation

\begin{aligned} \max_{w} \quad & \sum_{i=1}^{N} w_i \cdot S_i \\ \text{subject to} \quad & \sum_{i=1}^{N} w_i = 1 \\ & w_i \geq 0 \quad \forall i \\ & w_i \leq w_{max} \quad \forall i \end{aligned}

This LP formulation is solved using standard linear programming solvers, ensuring optimal allocation based on the strength of technical signals across the investment universe.

Advantages and Implementation Considerations

Advantages

Computational Efficiency: Linear programming is computationally faster and more stable than quadratic programming used in traditional MVO.
Robust to Estimation Errors: Avoids the need to estimate expected returns and covariance matrices, which are notoriously difficult to predict.
Cross-Sectional Focus: Captures relative strength patterns that are often more persistent than absolute return predictions.
Multiple Signal Integration: Systematically combines multiple technical indicators to reduce noise and false signals.
Adaptability: Framework can easily incorporate new technical indicators or adjust indicator weights based on market conditions.

Implementation Considerations

Look-ahead Bias: Must ensure all technical indicators use only historical data available at the time of portfolio construction.
Transaction Costs: High signal turnover may lead to excessive trading costs; consider implementing turnover constraints.
Market Regime Changes: Technical signals may lose effectiveness during structural market changes or unusual market conditions.
Overfitting Risk: Using too many indicators or complex combinations may lead to overfitting to historical data.
Beta Considerations: Unlike traditional optimization, this approach doesn't explicitly control for market beta or factor exposures.

Performance Characteristics for Indian Markets

Technical Indicator Optimization has shown particular effectiveness in Indian equity markets due to several unique characteristics:

Market Microstructure Benefits

Momentum Persistence: Indian markets exhibit stronger momentum effects compared to developed markets, making technical signals more predictive.
Cross-Sectional Dispersion: High dispersion in stock performance within sectors creates opportunities for relative strength strategies.
Retail Participation: Significant retail investor participation leads to behavioral patterns that technical indicators can effectively capture.

Risk Management Features

Adaptive Allocation: Weights adjust automatically based on changing signal strength, providing dynamic risk management.
Diversification: Maximum weight constraints ensure no single position dominates the portfolio.
Drawdown Control: Quick signal adaptation helps reduce portfolio drawdowns during adverse market conditions.

References

Jegadeesh, N., & Titman, S. (1993). "Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency."The Journal of Finance, 48(1), 65-91.Access the paper
Lo, A. W., Mamaysky, H., & Wang, J. (2000). "Foundations of Technical Analysis: Computational Algorithms, Statistical Inference, and Empirical Implementation."The Journal of Finance, 55(4), 1705-1765.
Park, C. H., & Irwin, S. H. (2007). "What Do We Know About the Profitability of Technical Analysis?"Journal of Economic Surveys, 21(4), 786-826.
Moskowitz, T. J., Ooi, Y. H., & Pedersen, L. H. (2012). "Time Series Momentum."Journal of Financial Economics, 104(2), 228-250.
Hurst, B., Ooi, Y. H., & Pedersen, L. H. (2013). "Demystifying Managed Futures."Journal of Investment Management, 11(3), 42-58.
Balvers, R., & Wu, Y. (2006). "Momentum and Mean Reversion Across National Equity Markets."Journal of Empirical Finance, 13(1), 24-48.
Wilder, J. W. (1978). New Concepts in Technical Trading Systems. Trend Research.
Bollinger, J. (2001). Bollinger on Bollinger Bands. McGraw-Hill Education.
Alvior, E. B. (2021). "Moving Average Indicator and Trade Set-up as Correlates to Investment Trading in Stock Market."Journal of Business and Management Studies, 3(2), 140-151.
Belantari, A. (2024). "An Optimization Enhanced Technical Momentum Method: Part One-Equities Portfolio."Medium Technical Analysis Research.

Technical Indicator Optimization
NEW

Cross-sectional signal optimization using technical indicators